CodeRabbit Data Handling

Overview

Manage code review data and sensitive patterns with CodeRabbit. Covers secret detection in PRs, sensitive file exclusion from AI review, review comment data retention, and configuring what code context gets sent to the AI engine.

Prerequisites

• CodeRabbit installed on repository
• Understanding of sensitive file patterns
• Repository admin access for configuration
• Secret scanning tools awareness

Instructions

Step 1: Exclude Sensitive Files from Review

# .coderabbit.yaml - Data handling configuration
reviews:
  path_filters:
    # Never send these to AI review
    - "!**/.env*"
    - "!**/credentials*"
    - "!**/secrets*"
    - "!**/*.pem"
    - "!**/*.key"
    - "!**/*.p12"
    - "!**/serviceAccountKey*"
    - "!**/terraform.tfstate*"
    - "!**/*.tfvars"

    # Exclude large generated files
    - "!**/package-lock.json"
    - "!**/pnpm-lock.yaml"
    - "!**/yarn.lock"
    - "!**/*.generated.*"
    - "!**/dist/**"
    - "!**/coverage/**"

Step 2: Secret Detection Instructions

# .coderabbit.yaml - Instruct AI to flag secrets
reviews:
  path_instructions:
    - path: "**"
      instructions: |
        CRITICAL: Flag any of these patterns as HIGH SEVERITY:
        - Hardcoded API keys, tokens, or passwords
        - AWS access keys (AKIA...)
        - Private keys or certificates
        - Database connection strings with credentials
        - JWT secrets or signing keys
        - Webhook URLs with tokens in query params

        If you find any secrets, add a comment:
        "SECURITY: Hardcoded secret detected. Move to environment variable."

    - path: "**/*.{yml,yaml}"
      instructions: |
        Check CI/CD files for:
        - Secrets logged in step names or echo statements
        - Unpinned GitHub Actions (use SHA, not tags)
        - Missing secret masking in outputs

Step 3: Review Data Scope Management

# Control what context CodeRabbit accesses
reviews:
  auto_review:
    enabled: true
    drafts: false   # Don't review draft PRs (may contain WIP secrets)
    base_branches:
      - "main"
      - "develop"
    ignore_title_keywords:
      - "WIP"
      - "DO NOT REVIEW"
      - "DRAFT"

  # Limit file types reviewed
  path_filters:
    # Only review source code, not data
    - "+src/**"
    - "+lib/**"
    - "+app/**"
    - "+tests/**"
    - "+.github/**"
    - "!**/*.csv"
    - "!**/*.json"      # Exclude data files
    - "!**/fixtures/**"  # Exclude test fixtures with sample data
    - "!**/seeds/**"     # Exclude database seeds